Posted by Jeff C from 192.156.168.6 (192.156.168.6)on March 09, 1998 at 12:52:06:
I spent a fair bit of time last night updating some more things.
Turns out that Glide sets up a whole lot of stuff for color in
grColorCombineFunc() . I am trying to get the wrapper to detect
when NOT to use the RGB values that are in the vertex structure,
and it does not seem to be going well.
The main thing is that I cannot debug the apps under VC++. When they go
full screen, they really mean it!! If I set a break point, I just still
see the full graphics screen. I think I might have to try to see If I can
get this thing to render in a window so that I can debug it.
Also related, I can't seem to get the grConstantColor() thing to work right.
As I understand it, if you send GR_LOCAL_COLOR to grColorCombineFunc, you
won't get shaded surfaces, you will get everything one color, the constant
color. Well, I can detect this situation, but I can't seem to tell redline
that it should be drawing in a certain color. I think I might have to set
up a different color function in redline to do it.
(I am at work, and I left all the code at home, so I might have typed some wrong
function names or constants above)
I implemented fog last night too, but i am not sure it works. I think it has something
to do with the color thing I talked about earlier.
Glide has a concept of a fog table so you can have essentially a custom fog generation
function. This will never be implemented on the current verite chips.
I started to think about texturing though. Unfortunately, most of the smaple programs
use guLoad3df. This is bad. i don't think anyone knows how a .3df file is written, much
less read. if someone can find a program that writes out .3df files (with source code),
all of us working on this project would be much obliged.
I would also be thankful if we could get someone who has a 3dfx card to take screenshots of
each of the demo programs so that we can see what they look like on 3dfx hardware.
Does anyone know someone with a 3dfx card that would do it?
Last thing:
I am starting to reach a point where I might start to worry about alpha values. The project currently
ignores alpha. I am thinking of changing the vertex structure to be
V_PIXFMT_8888 instead of V_PIXFMT_565.
Anybody have any objections to this? I know it might be slower, but we really aren't worried about
slowing down the simple applications, we want to speed up the complex applications.
Besides, I still want to set it up so that we can convert to some different types of vertices, rather
than just one.
No binary update because nothing about the external appearance significantly changed, but there are
quite a few source code updates. If you want, mail me, and i can send you the new source.